• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2001.06a
• /
• pp.285-288
• /
• 2001

In this paper, we proposed new method of speech feature extraction using ICA(Independent Component Analysis) which minimized the dependency and correlation among speech signals on purpose to separate each component in the speech signal. ICA removes the repeating of data after finding the axis direction which has the greatest variance in input dimension. We verified improvement of speech recognition ability with training and recognition experiments when ICA compared with conventional mel-cepstrum features using HMM. Also, we can see that ICA dealt with the situation of recognition ability decline that is caused by environmental noise.

• Smart Media Journal
• /
• v.6 no.3
• /
• pp.57-63
• /
• 2017

The most important, essential prerequisite for the improvement of vehicle license plate recognition is the acquisition of high-quality vehicle images. Because typical images acquired from roads are affected by different environmental factors including the time of day, sunlight, and the weather, the brightness and the shape of the license plates in the images are inconsistent. To this end, many image corrections are performed, resulting in slower recognition and lower recognition rate. Therefore, in this study, we used the images acquired from roads to test the proposed method for fast capturing of vivid, high-quality vehicle images by measuring the brightness around license plates during real-time image capturing to control in real time the factors, such as shutter speed, brightness, and gain of the camera, that affect the brightness and the quality of the images.

• Journal of IKEEE
• /
• v.24 no.4
• /
• pp.1086-1092
• /
• 2020

An autonomous agent for real world should be able to recognize text in scenes. With the advancement of deep learning, various DNN models have been utilized for transformation, feature extraction, and predictions. However, the existing state-of-the art STR (Scene Text Recognition) engines do not achieve the performance required for real world applications. In this paper, we introduce a performance-improvement method through an add-on composed of an OCR (Optical Character Recognition) engine and a classifier for STR engines. On instances from IC13 and IC15 datasets which a STR engine failed to recognize, our method recognizes 10.92% of unrecognized characters.

• Journal of the Semiconductor & Display Technology
• /
• v.21 no.3
• /
• pp.50-56
• /
• 2022

In order to process a vast amount of data, there is demand for a new system with higher processing speed and lower energy consumption. To prevent 'memory wall' in von Neumann architecture, RRAM, which is a neuromorphic device, has been researched. In this paper, we summarize the features of RRAM and propose the device structure for characteristic improvement. RRAM operates as a synapse device using a change of resistance. In general, the resistance characteristics of RRAM are nonlinear and random. As synapse device, linearity and uniformity improvement of RRAM is important to improve learning recognition rate because high linearity and uniformity characteristics can achieve high recognition rate. There are many method, such as TEL, barrier layer, NC, high oxidation properties, to improve linearity and uniformity. We proposed a new device structure of TiN/Al doped TaO_x/AlO_x/Pt that will achieve high recognition rate. Also, with simulation, we prove that the improved properties show a high learning recognition rate.

• Phonetics and Speech Sciences
• /
• v.10 no.1
• /
• pp.33-38
• /
• 2018

Although deep neural network-based acoustic models have greatly improved the performance of automatic speech recognition (ASR), reverberation still degrades the performance of distant speech recognition in indoor environments. In this paper, we adopt the DenseNet, which has shown great performance results in image classification tasks, to improve the performance of reverberant speech recognition. The DenseNet enables the deep convolutional neural network (CNN) to be effectively trained by concatenating feature maps in each convolutional layer. In addition, we extend the concept of multi-resolution CNN to multi-resolution DenseNet for robust speech recognition in reverberant environments. We evaluate the performance of reverberant speech recognition on the single-channel ASR task in reverberant voice enhancement and recognition benchmark (REVERB) challenge 2014. According to the experimental results, the DenseNet-based acoustic models show better performance than do the conventional CNN-based ones, and the multi-resolution DenseNet provides additional performance improvement.

• International journal of advanced smart convergence
• /
• v.9 no.1
• /
• pp.193-201
• /
• 2020

The main objective of this study is to investigate the impact of additional modalities on the performance of emotion recognition using speech, facial expression and physiological measurements. In order to compare different approaches, we designed a feature-based recognition system as a benchmark which carries out linear supervised classification followed by the leave-one-out cross-validation. For the classification of four emotions, it turned out that bimodal fusion in our experiment improves recognition accuracy of unimodal approach, while the performance of trimodal fusion varies strongly depending on the individual. Furthermore, we experienced extremely high disparity between single class recognition rates, while we could not observe a best performing single modality in our experiment. Based on these observations, we developed a novel fusion method, called parametric decision fusion (PDF), which lies in building emotion-specific classifiers and exploits advantage of a parametrized decision process. By using the PDF scheme we achieved 16% improvement in accuracy of subject-dependent recognition and 10% for subject-independent recognition compared to the best unimodal results.

• Journal of Korea Multimedia Society
• /
• v.16 no.8
• /
• pp.905-913
• /
• 2013

The Parallel Model Compensation Using Frequency-variant: FV-PMC for noise-robust speech recognition is a method to classify the noises, which are expected to be intermixed with input speech when recognized, into several groups of noises by setting average frequency variant as a threshold value; and to recognize the noises depending on the classified groups. This demonstrates the excellent performance considering noisy speech categorized as good using the standard threshold value. However, it also holds a problem to decrease the average speech recognition rate with regard to unclassified noisy speech, for it conducts the process of speech recognition, combined with noiseless model as in the existing PMC. To solve this problem, this paper suggests a enhanced method of recognition to prevent the unclassified through improving the extent of rating scales with use of maximum likelihood so that the noise groups, including input noisy speech, can be classified into more specific groups, which leads to improvement of the recognition rate. The findings from recognition experiments using Aurora 2.0 database showed the improved results compared with those from the method of the previous FV-PMC.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.11 no.8
• /
• pp.1596-1603
• /
• 2007

In this paper, we experimented by three kind of method for feature parameter, training method and recognition algorithm of most suitable for speech recognition system and considered. We decided speech recognition system of most suitable through two kind of experiment after we make speech recognizer. First, we did an experiment about three kind of feature parameter to evaluate recognition performance of it in speech recognizer using existent MFCC and MFCC new feature parameter that change characteristic space using PCA and ICA. Second, we experimented recognition performance or HMM, SVM and RVM by studying data number. By an experiment until now, feature parameter by ICA showed performance improvement of average 1.5% than MFCC by high linear discrimination from characteristic space. RVM showed performance improvement of maximum 3.25% than HMM in an experiment by decrease of studying data. As such result, effective method for speech recognition system to propose in this paper derives feature parameters using ICA and un recognition using RVM.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.2
• /
• pp.318-328
• /
• 2014

In this paper, as a way of performance improvement of business card recognition in the mobile environment, we suggested a hybrid OCR agent which combines data using a parallel processing sequence between various algorithms and different kinds of business card recognition engines which have learning data. We also suggested an Image Processing Method on mobile cameras which adapts to the changes of the lighting, exposing axis and the backgrounds of the cards which occur depending on the photographic conditions. In case a hybrid OCR agent is composed by the method suggested above, the average recognition rate of Korean business cards has improved from 90.69% to 95.5% compared to the cases where a single engine is used. By using the Image Processing Method, the image capacity has decreased to the average of 50%, and the recognition has improved from 83% to 92.48% showing 9.4% improvement.

• Korean Journal of Community Nutrition
• /
• v.24 no.4
• /
• pp.290-299
• /
• 2019

Objectives: This study examined the nutrition teachers' and parents' recognition of environmentally-friendly agricultural products (EAPs) used in school foodservice. Methods: A questionnaire survey was given to 128 school foodservice nutrition teachers in Seongnam and 189 parents from Oct. 16 to Oct. 31, 2018 at Seongnam in Gyeonggi province. The survey included information on the recognition, satisfaction, and improvement of EAPs, and the results of the two groups were compared. Results: A comparison of the recognition of EAPs showed that nutrition teachers knew more about the EAPs and local government support in school foodservice than the parents. On the other hand, the parents were more aware than the nutrition teachers in that children have a higher affinity for EAPs than for general agricultural products in the school foodservice. A comparison of the level of satisfaction with the EAPs by nutrition teachers and parents revealed the nutrition teachers to be significantly more satisfied than parents in terms of the color, taste and nutrition of EAPs. Among the items that should be provided with EAPs, more than 50% of each group of nutrition teachers and parents answered that vegetables must be provided first. Some 70.9% of nutrition teachers and 84.5% of parents were aware of the certification standards of EAPs. The nutrition teachers had showed a slightly higher score than the parents in the certification system (3.51 vs. 3.25). In terms of improving the EAPs, 36.2% of nutrition teachers answered a reasonable price preferentially, whereas 56.4% of parents answered maintaining quality. In the expected effects of using EAPs, 57.9% of nutrition teachers answered an improvement of parents' satisfaction on the school foodservice. On the other hand, 38.0% of parents answered an improvement of children' satisfaction on school foodservice. Conclusions: Nutrition teachers and parents need to be educated on the certification systems that would enhance the trust in EAPs.